Unexplained flip and crash in auto mode

Hi all,

While doing an auto flight at 400 ft yesterday in 7mph winds, the craft took a tumble! Remarkably, it landed gently in a tree with only two props broken.

Looking at the logs, I can't find any anomalies. I thought maybe it's was hardware, but the craft flies totally fine, so I took to the web. 

I see other people who've had this problem tend to blame a motor failure or esc brown out/failure. Looking at my logs, when it flipped it did it almost perfectly along the roll axis as shown in the logs.

3691110023?profile=original

Since my craft is an "x" config quad, it means both the right esc's would have to brown out at the same time, as the motors are still fully operational, OR all four esc's browned out and it tumbled that direction by chance.

All four browning out seems unlikely since the roll was so strong and quick, so ruling that out, I'm left with that two esc's browned out at the same time. 

Is this possible? I wouldn't think so, since all 4 esc's get their power on the same parallel circuit. So is there any other explanation?

As you can see, after 2.5 flips and just over 100 ft I flipped to "loiter" mode. Unfortunately, the craft managed to stabilize inverted! I know arducopter sends no signals to motors when it's upside down, so if that's an inverted aerodynamic stabilization, it happened VERY quickly after switching to "loiter." This would imply that during the flips, some props were spinning to help drive the roll.

EDIT: Looked at servo outs. Three of the motors were commanded to spin up right when the stabilize command happened. Just a command though. Doesn't mean they did I guess.

Here are my log files for this. Can anyone help me figure out what happened here? It's a weird one!

Also, do we not have any sort of automatic "flip recover" on the arducopter?! I know in the flip command there are 3 steps, start, roll, and recover. Can this not automatically be applied when upside down in certain "stabilize" and "loiter" modes?

Thanks for any help you can give!

-Colin

2014-03-31 13-17-14.tlog

2014-03-31 13-19-33.log

You need to be a member of diydrones to add comments!

Join diydrones

Email me when people reply –

Replies

  • Oh no! It's happened again!

    Checked all mechanical connections on the craft with everything looking good. I flew it 5 ft off of the ground in a gentle back and forth covering about 10 ft horizontally for a full battery in loiter mode. Then I gave it a full autonomous flight at 200 ft and everything was looking great!

    On the second, fully autonomous flight right as it switched to waypoint 6, it rolled to the left again. This time I was able to switch it to loiter much quicker so it only made one revolution before the craft caught itself. I RTL and if flew uneventfully home to a safe landing. 

    What's going on with this?! I've done these autonomous flights at least 20 times and never seen this behavior. I don't think I updated the arducopter code recently at all. It's pretty frustrating as it makes my ardupilot useless!

    Here are the logs today. I thought I told it to log motor stuff too, but apparently I failed to do that. Just the standard flashlog I'm afraid.

    Any help would be GREATLY appreciated! 

    Thanks,

    -Colin

    2014-04-07 13-17-58.log

    2014-04-07 13-11-46.tlog

    • WOW, sorry to hear that.  The good news is that it seems to be tied to Mission Planner if it truly only happens in Auto mode.    Was it the same mission or a new one?

      I have not looked at the logs yet, but I would strongly suggest you do an erase/reset of your eeprom in the CLI.... mission planner.   You'll have to recalibrate the compass, the accel and the radio but it is worth it.

      The only other issue I would suspect is something to so with the telemetry radio like a loose wire on the connector.  Take a good look at those.

      Also, have you checked the PWM value for the AUTO mode to see if it is really close to the center of the band for that mode?

      Last, but not as likely would be interference from something like a video transmitter, but if you have used this set up many times before without adding anything that is not as likely...

      I have only been flying quads for a year so I don't have a lot of experience.   Just trying to help.

      • 3702646562?profile=originalColin,

        I did look at the logs.   The only thing I see a bit weird, is the current drops quite hard just before you switched to loiter.   see it in this screen shot?   See it also happens just before the roll begins.  Not sure what to make of it....but...it is there.  

        enable the MOTORS and the IMU and the COMPASS (or is it MAG, I cannot remember???)   in the CLI....

  • Ah, it should be noted that this was a relatively new craft. I've flown about 45 hours with the exact same set up, so I got a second one. This is the second one. It's been through my rigorous flight test regimen and completed 4 auto flights before this happened.

    • 3702482860?profile=originalI have limited knowledge on this stuff....so my thoughts should not be taken too seriously, but, I did look through your logs and I did not see anything that stood out to me.   I would have liked to see the IMU values in the logs and MOTORS and MAG.      There is a compass interference value in the MAG logs called MOfsx  that could potentially reveal any motor interference with the mags...

      Are you using an external mag?

      Have you checked your compassmot value?

      I will try to attach a screen shot of what the mofsx value looks compared to the roll  values.   They should be somewhat in sync.  I was doing some seriously hard rolls in this graph.   If you were to see high mofsx with no hard x,y,z inputs then that would be a potential issue revealing some interference..

      • Thanks for checking! I just started reconfiguring my craft a bit today, so I can't check my compassmot values just yet, but will do, thanks. I do have the external mag on the 3dr gps/mag so hopefully interference isn't the issue!

        I'll check and post as soon as my craft is back together. Thanks!

        -Colin

        • I ran compassmot and it says 21% when I was just above cruise throttle. Looks like that's well within the suggested values. 

          • Good to know.     If you could add the motors to the logs that may help see if there was some anomaly with one of them that caused it in the future.   Like a loose connection or a sporatic value from the esc  etc.   Last time this happened to me I found a loose connection of one of my motor wires.  

            Good flying to you.

This reply was deleted.

Activity